27-10-2016, 11:30 AM
1461610750-FinalProject1.pdf (Size: 505.33 KB / Downloads: 7)
Abstract- Advancements in the field of
healthcare has enabled the use of wearable
and implantable devices to improve the
quality of treatment for a variety of medical
applications. However as these wearable
devices are wirelessly connected, there is a
huge opportunity for malicious attacker to
cause harm to the patients whose health is
being monitored. Cryptographic techniques
are often difficult to be implemented on
these devices due to their resource
constraints. In this paper, we harness the
unique characteristics of Wireless Medical
Sensor Networks (WMSNs). This technique
is based on trust and cooperative nature of
WMSNs. This trust management system
evaluates trust by taking into account stable
transmission rates of sensors in WMSNs. In
addition to this, mobile agents are employed
to distribute and evaluate trust among the
sensor nodes and mobile gateway in the
WMSN. Simulation results reveal that
proposed model can identify node
misbehavior based on transmission rate of
sensor nodes and thereby exclude malicious
nodes in the network. Experimental results
also reveal that changes in distance of the
malicious sensor node from the base station
also causes significant effect on the detection
rate.
Introduction
WMSNs have great potential in realtime
monitoring of patients in the field of
healthcare and medicine [1]. With the promise
of cost effective, automatic and intelligent
supervision, WMSNs have attracted a wide
range of applications from sports to healthcare.
Though many of the key issues in healthcare industry is resolved by this technique, there are
still existing concerns of privacy and security
with the widespread usage of WMSN. As these
wearable systems have the ability to handle life
critical applications, it is necessary that these
systems can secure sensitive data. Often,
addressing these security issues with existing
solutions face difficulties in case of WMSNs
[2]. There are several cryptographic techniques
in place to protect private data of patients. But
the bigger problem is the heaviness of
cryptographic algorithms that could not be
effectively employed on the resource
constrained medical sensor nodes. Moving in
this direction, this paper primarily focusses on
building a trust management system for
WMSNs.
Trust evaluation techniques are
extensively used for applications like wireless
sensor networks, vehicular ad-hoc networks,
mobile ad-hoc networks (MANET) [3].
However, existing trust management
techniques are limited to MANETs [4] and
could not be extended to WMSNs due to the
unique features of medical sensors. Hence, it is
necessary to establish trust between different
entities in the medical sensor network to
safeguard the patient’s medical data.
Related work
Although there are existing reputation
based trust mechanisms that identifies
malicious node activity but cannot effectively
revoke the malicious node. However, these
techniques require the sensors to be monitored
periodically. One of the simple trust
management technique TrE [5] for medical
sensor network is based on secure multicast
which could provide better packet delivery when compared to existing techniques but fails
to provide an account of failed packets. This
technique is also vulnerable to collaborative
attacks.
In ReTrust [6], Daojing et al. proposed
a two tier trust management technique. ReTrust
is a trust mechanism which can fight against
trust attacks. Trust worthiness of a node is
evaluated from direct, indirect and
recommended trusts. According to ReTrust
model, MSN basically is divided into three
essential components- Sensor node (SN) and
Master node (MN) and Base Station (BS).
Though this technique could handle
intercellular and intracellular attacks, the
influencing factors like trust threshold,
forgetting factor and time window are not
clearly defined. Though it could overcome badmouthing
and ballot stuffing attack, but if the
trust threshold is maintained, a malicious node
can survive in the network thus making it
impossible to detect an adversary in the
network.
To mitigate the limitations imposed by
these existing methods, this project proposes a
mobile agent based approach to derive at the
optimal strategies for detecting the
compromised node as well as reducing false
positives, by identifying the malicious node in
the network. The solution provided is a novel
approach of detecting a compromised node by
building an effective trust management system
by injecting mobile agents into the network.
Working of Mobile Agents
Mobile agents are widely used in
distributed systems for data retrieval processes.
Network users launch mobile agents to
accomplish a particular task while migrating
from one node to another. Mobile agents are
robust and adaptable to a wide range of system
conditions. A mobile agent contains data and
code which is executable at a place by sharing
the resources available on these sensor nodes.
It can also accommodate to the changes happening in the network. So, this makes the
mobile agent facilitate a self-healing process
which has its own architecture.
Network Model
Fig.1 represents a distributed
architecture of the medical sensor network. The
sensor network is divided into number of zones
where each zone consists of a mobile gateway
and set of sensor nodes reporting to this mobile
gateway. Thus, the mobile gateways are
entrusted with the responsibility of evaluating
the trust of the sensor nodes. This helps in
disintegrating the network into separate zones
which performs trust calculation. Each mobile
gateway has an agent to accumulate the trust of
sensor nodes and other sensor nodes. Hence,
this model is categorized into two types of
nodes:
1. Mobile gateways: A mobile gateway
receives the data from the sensor node
and forwards this data to the base
station in order to perform certain
operation on the data. It also evaluates
the trust of sensor nodes in a particular
zone. In Fig. 1, a WMSN network is
divided into six zones. Agent injector
resides inside each mobile gateway in a
zone.
2. Sensor nodes: It senses a specific action
and forwards this data to base station
through the mobile gateway.
3. Mobile agents: Mobile agents are
injected into the network [7]. There are
two agents injected into the system:
Local agents (Lagmo) and Global agent
(Gagmo). Local agent maintains the
trust table within a zone i.e. between the
sensor nodes and mobile gateway in a
zone. Consequently, a local agent
moves in each zone. Global agent
maintains the trust values among different mobile gateways in the
network. Thus, this agent goes around
the entire network.
Assumptions
Mobile agents are injected into the
network through an agent injector
Overview of mobile agent based trust
model
In order to compute and manage trust
values, this model uses mobile agents [7] to
administer the trust values as medical sensor
nodes will not be able to calculate and
manage their own trust values due to energy
constraints. In this model, local mobile
agents manages one-to-one trust between the
sensor nodes and the mobile gateway while
global agents maintains trust between the
various mobile gateways in the network.
From Fig. 2, consider a WMSN
network divided into a number of zones
where each zone is made up of a mobile
gateway and a set of sensor nodes [6]. In an ideal case scenario, mobile gateway
evaluates and manages the trust of sensor
node. This causes a resource and energy
overhead on the mobile gateway. In addition
to it, mobile gateway only has the ability to
see if a sensor node is trustworthy and this
data is stored on the mobile gateway. There
is a lack of mutual trust evaluation between
the sensor node and mobile gateway.
Consider a situation where mobile gateway is
malicious and extracts data from legitimate
sensor nodes, it could open up a huge security
threat to the entire network. A solution to this
problem is injecting a mobile agent [7] into
the network. As a result, once the mobile
gateway evaluates the trustworthiness of a
sensor node. This trust is calculated and stored in the local mobile agent 1 (Lagmo1).
Lagmo1 is then sent to sensor node “a”.
Inside the sensor node, Lagmo1 computes
and stores the trust value of the mobile
gateway A. This helps in mutual cross
evaluation of trust on master and sensor node.
Lagmo1 is again sent to the mobile gateway
“A”. In mobile gateway “A”, Lagmo1 begins
to evaluate trustworthiness of sensor node
“b” and the process continues till the trust of
the zone is gauged.
On the other hand, the global agent is
launched into the network from the agent injector. The global agent (Gagmo) manages
the trust value of all the mobile gateways in
the network. It moves from one mobile
gateway to another. From Fig.2, it can be
seen that at mobile gateway “A”, Gagmo
evaluates and stores the trust of mobile
gateway “B” who is the immediate neighbor
of “A”. Then Gagmo moves towards node
“B”. From “B”, global mobile agent Gagmo
also evaluates the trust of all its one hop
neighbors “C”,”D” and then moves to the
respective nodes to oversee the trust of the
other mobile gateways.
Conclusion
In this paper, a novel mobile agent
based trust management system for WMSNs
is being proposed, where mobile agent
manages the trust records of other nodes
based on the transmission rate. Furthermore,
this paper reveals that this model can be
implemented in a variety of application
scenarios in WMSNs. The proposed model is
simulated using Castalia software to detect
node misbehaviors and helps in routing the
data through the most trusted path instead of
shortest path.